A SIMULATION MODEL FOR MANAGING ENGINEERING
CHANGES ALONG WITH NEW PRODUCT DEVELOPMENT
Weilin Li and Young B. Moon
L. C .Smith College of Engineering & Computer Science, Syraceuse University, 223 Link Hall, Syracuse, NY 13244, U.S.A.
Keywords: Engineering Change Management, New Product Development, Process Modeling, Discrete-Event
Simulation.
Abstract: This paper presents a process model for managing Engineering Changes (ECs) while other New Product
Development (NPD) activities are being carried out in a company. The discrete-event simulation model
incorporates Engineering Change Management (ECM) into an NPD environment by allowing ECs to share
resources with regular NPD activities. Six model variables - (i) overlapping, (ii) NPD departmental
interaction, (iii) ECM effort, (iv) resource constraints, (v) arrival rate, and (vi) resource using priority - are
explored to identify how they affect lead time and productivity of both NPD and ECM. Decision-making
suggestions for minimum EC impact are then drawn from an overall enterprise system level perspective
based on the simulation results.
1 INTRODUCTION
Today’s increasingly competitive market forces any
corporations who develop new products to look into
all the possible areas of improvement in their entire
product lifecycle management process. One of the
areas that have been overlooked in the past is the
Engineering Change Management. Engineering
Change Management (ECM) refers to a collection of
procedures, tools, and guidelines for handling
modifications and changes to a product after the
product has been released to the market. (Terwiesch
and Loch, 1999; Bhuiyan, 2006) In reality, an ECM
is a norm rather than an exception in any typical
product development firm. Consequently, ECM
plays a critical role in finally realizing actual profits
from new product development efforts.
While the demand for more effective ECM has
increased, managing EC also became more efficient
than ever due to various advancements in tools and
technologies. The digitalized virtual design and
prototyping tools provide greatly increased
efficiency but with shorter cycle time and less cost.
The integrated Enterprise Resource Planning system
(Moon, 2007) assists ECM by eliminating redundant
documentation, assuring data consistency, and
maximizing data sharing among affected parties.
The main question is how to bring these new
available aids to enhance the ultimate new product
development process. Particularly, we are interested
in investigating how ECM affects general New
Product Development (NPD) activities and vice
versa.
Most of previous researches studying the ECM
processes focused on general administrative rules for
an organization to follow to reduce long lead-time of
ECMs, regardless of the types of firms and products
or other diverse operational conditions. (Wright,
1997) However, the following several important
issues impact both NPD and ECM significantly:
First, product development firms of different
sizes design and manufacture products differently
with varying degrees of complexity. In other words,
the frequency of developing new products and that
of handling engineering changes can be quite
different from one company to another. Also in
general, engineering change requests (ECRs) occur
in far more random patterns. Second, ECRs that
require modification or rework in different NPD
stages need different amount of time and effort.
Third, firms may choose to employ different
structure for its NPD process depending on how they
handle coupled product development activities and
cross-functional interaction among departments.
Fourth, NPD and ECM activities normally compete
for limited resources available in a firm. Therefore,
13
Li W. and B. Moon Y. (2009).
A SIMULATION MODEL FOR MANAGING ENGINEERING CHANGES ALONG WITH NEW PRODUCT DEVELOPMENT.
In Proceedings of the 11th International Conference on Enterprise Information Systems - Information Systems Analysis and Specification, pages 13-18
DOI: 10.5220/0001852200130018
Copyright
c
SciTePress
firms have to allocate available resources between
NPD and ECM to maximze their ultimate profits.
Weighing the above four major factors,
companies may adopt different NPD and ECM
strategies. The objective of this research is to
provide insightful decision-making suggestions for
companies regarding how engineering changes
should be implemented with minimal adverse effects
on normal NPD activities. We propose to model and
simulate the EC implementation within a multi-
project development environment to answer the
following questions:
1. How important is ECM for a firm that is
engaged in developing new products?
2. What are the key contributors to long lead
times in NPD in relation with ECM? And
vice versa.
3. What are the key contributors to low
production rates in NPD in relation with
ECM? And vice versa.
4. What is an optimal way of allocating limited
resources between NPD and ECM?
A discrete-event simulation methodology is
adpoted to model both NPD and ECM process
together, primarily because of their complexity.
2 LITERATURE REVIEW
The review of papers until 1995 was done by
Wright. (Wright, 1997) The author categorized the
EC related papers into two main topics, computer-
based “tools” for the analysis of EC problems and
“methods” to reduce the impact of ECs on
manufacturing and inventory control. We can find
that most of the publications in that time period
predominantly focused on the EC administrative
guidelines and control mechanisms. An important
observation by Wright is that understanding of the
positive effect EC can provide for product
improvement and enhanced market performance is
long omitted by EC research.
Terwiesch and Loch presented a process-based
view of ECM. (Terwiesch and Loch, 1999) They
showed by an industrial case study that a
complicated and congested administrative support
process is one of the root causes of long lead time
and high cost. Based on the field study, they
indentified five key contributors to lengthy ECO
lead time: complex ECO approval process, scarce
capacity and congestions, setups and batching,
snowballing changes, and organizational issues.
In another paper they wrote, an analytical
framework that explains the extreme ratio between
theoretical processing time and actual lead time was
developed. (Loch and Terwiesch, 1999) They
showed how congestion and batching influence
engineering processes at a more detailed level.
Based on the processing network framework, they
suggested improvement strategies such as flexible
work times, the grouping of several tasks, workload
balancing, the pooling of resources, and the
reduction of setup times.
Krishnan (Krishnan, 1997) presented a model-
based framework to manage the overlapping of
coupled product development activities. The author
introduced two properties, upstream information and
downstream iteration sensitivity, of the information
exchanged between product design phases. The
mathematical model and conceptual framework of
the overlapped process were illustrated with
industrial examples to provide managerial insights.
Bhuiyan and her co-workers built a stochastic
computer model to examine how overlapping and
functional interaction affect the performance
measures of development time and effort under
varying conditions of uncertainty. (Bhuiyan, 2004) It
is the first comprehensive model using a discrete-
event simulation for the entire NPD process by
taking into account functional interaction at different
values of overlapping under different uncertainty
conditions. Development effort was also introduced,
in the form of total person-days for a project, as a
measure of NPD performance that was neglected by
earlier researchers. A number of conclusions were
drawn from the model, however, their model
assumed an unlimited amount of resources, which is
unrealistic in practice.
Bhuiyan’s research group has also expanded this
framework to compare two methods for managing
Engineering Change Requests (ECRs): immediate
individual processing as issued and batch processing
after accumulation. (Bhuiyan, 2006) They evaluated
the effects of the methods in terms of development
time and effort. The model they developed, though,
has a couple of limitations: (i) the research scope
only on immediate or batch processing, is too
simplified compared with a large amount of ECM
problems; (ii) treating all ECRs similarly is
acceptable only for comparative analysis. Despite of
these limitations, Bhuiyan’s model is the only study
on ECM using the discrete-event simulation. Thus it
inspired our model.
Browning presented a thorough literature survey
on the topic of activity network-based models for
NPD project management. (Browning, 2007) The
paper is based upon four major categories:
visualization, planning, execution and control, and
ICEIS 2009 - International Conference on Enterprise Information Systems
14
project development. The author highlighted the
models’ main assumptions, findings, and insights.
To conclude, he identified five research directions
for future study: activity interactions, global process
improvements, process models as an organizing
structure for knowledge management, modeling in
cases of uncertainty and ambiguity, and determining
the optimum amount of process prescription and
structure for an innovative project.
3 MODEL OF NPD AND ECM
In this section, we will introduce the framework
structure, components, variables, parameter setting,
and assumptions of this modeling project. Arena
simulation package is used for the project.
NPD Framework
The NPD model has three phases, namely Concept,
Design, and Production. They occur sequentially but
with certain degrees of overlapping. Each phase is
consisted of three sequentially numbered activities
to represent its different stages.
Assumptions
The model assumptions are presented in the bullet
form [A1], [A2], ...[An].
[A1] Each NPD project begins with an inter-
arrival value of 20 days, 48 days, or 120 days
depending on project size and product type.
Correspondingly, the arrival rate expressed in Arena
is CONST 12/yr, 5/yr, and 2/yr.
[A2] The activity duration follows a normal
distribution, which represents the uncertainties in
product design and development processes.
[A3] The mean value of activity duration within
one phase remains the same, but increases as NPD
entities proceed from Concept to Design to
Production because of the increasing activity
complexity since more product development tasks
are involved. Detailed activity duration assignment
is shown in Table 1.
[A4] When NPD arrives at a lower (or higher)
rate, we assume the project to be more (or less)
complicated and thus require more (or less) time to
finish. The duration of an activity is set to be
proportional to its arrival rate.
NPD Overlapping
In this paper we refer to overlapping as the partial or
full parallel execution of tasks. By having this 3-
phase and 3-activity framework, we are able to
construct an NPD process with 0% (sequential),
33%, or 66% overlapping, while any amount
between 0 and 100% can be true in real life.
Table 1: NPD Activity Duration.
NPD
Arrival
Rate
NPD Activity Duration in
Concept
phase
Design
phase
Production
phase
CONST
12/yr
NORM
(1.333,
0.645)
NORM
(2, 0.791)
NORM
(3.333,
1.021)
CONST
5/yr
NORM
(3.2, 1)
NORM
(4.8,
1.225)
NORM
(8, 1.581)
CONST
2/yr
NORM
(8, 1.581)
NORM
(12, 1.936)
NORM
(20, 2.5)
NPD Iteration
After each activity, there is a Decision module in
which NPD entities pass through or go back by pre-
assigned probability. NPD entities may go back and
repeat the just-finished activity or any one of its
previous activities, including activities in other
phases. This rework process is called NPD iteration.
Probability of the N-way decision by chance to go
back to one certain activity for rework is also
modelled.
Departmental Interaction
The concept of cross-functional integration among
different functional areas during an NPD process is
defined as departmental interaction.
One of the three departments - Marketing,
Design, and Manufacturing - takes major
responsibility for a phase of its own specialization,
and is called major department during that phase. In
other words, Marketing Department is the major
department in Concept phase, Design Department in
Design phase, and Manufacturing Department in
Production phase. However, the other two
departments, defined as minor departments, also
participate in the same phase with less allocation of
resources.
Two levels of departmental interaction, 60
(major dept.) - 20 (minor dept.) - 20 (minor dept.),
and 40 (major dept.) - 30 (minor dept.) - 30 (minor
A SIMULATION MODEL FOR MANAGING ENGINEERING CHANGES ALONG WITH NEW PRODUCT
DEVELOPMENT
15
dept.), are examined in our model. These two levels
represent low and high departmental interaction with
a total resource consumption of 100 number of
resources.
Resource Constraints
Resources can represent staffs, computer/machine,
documentation support, or any other individual
server. In our model, we examined three levels
resources, that is, 200, 100, or 60 numbers of
resources from each department. We set the
minimum number to be 60 resources per department,
which is equal to the resource consumption for a
major department at low level of departmental
interaction, in order to ensure that the major
department gets enough resource to let NPD process
flow.
We assume that each resource is qualified to
handle all the NPD activities in three phases.
ECM Framework
Additional assumptions for ECM model include:
[A5] One EC is confined in only one NPD
activity in this model.
[A6] ECM shares the same pool of resources
with that particular NPD activity by defining its
queue as shared.
[A7] Concept 3, three activities in Design, and
three activities in Production each have an equal
chance of implementing an ECR.
[A8] Changes that are undertaken in Concept 1
and Concept 2 are not considered as ECs since
within the first two NPD activities a
comprehensively large number of new product ideas
are gathered, discussed and modified. NPD ideas are
less formally organized.
[A9] Compared with NPDs that are more likely
sticking to a planned schedule, ECRs occur without
expectations. So we use exponential distribution to
assume ECRs’ arrival.
[A10] The ECM process time is set to be
proportional to its corresponding arrival rate. It also
increases proportionally from phase to phase in the
same fashion as NPD activity duration does. Table 2
shows the detailed process time for an ECR to be
implemented within different NPD phases at three
arrival rates.
ECM Effort
The amount of resources required for an EC to be
processed is called ECM effort. Three levels of ECM
effort, 2-2-2, 5-5-5 and 10-10-10, are examined in
this model.
We assume that an EC consumes equal number
of resources from all three departments no matter in
which phase it occurs.
Table 2: ECM Process Time.
ECM
Arrival
Rate
ECM Process Time in
Concept
phase
Design
phase
Production
phase
Random
(EXPO)
8/mo
TRIA
(0.25, 0.5,
0.75)
TRIA
(0.38, 0.75,
1.12)
TRIA
(0.62, 1.25,
1.88)
Random
(EXPO)
4/mo
TRIA
(0.5, 1,
1.5)
TRIA
(0.75, 1.5,
2.25)
TRIA
(1.25, 2.5,
3.75)
Random
(EXPO)
2/mo
TRIA
(1, 2, 3)
TRIA
(1.5, 3, 4.5)
TRIA
(2.5, 5, 7.5)
Resource using Priority
When there are not enough resources available for
both processes, a priority needs to be assigned to
either NPD or ECM to get resource first.
This is achieved by setting priority to seize
resource in Process and Seize modules in Arena.
Running Parameters
We’ve specified the running parameters Hours-Per-
Day as 8 and Work-Day-Per-Year to be 20
days/month * 12 months/year (240 days/year). We
run the model in ten replications with a replication
length of 2 years.
4 RESULTS ANALYSIS
For the model described above, we analyzed the
influence of resource constraint, resource using
priority, overlapping, NPD departmental interaction,
ECM effort, on both NPD and ECM lead time and
productivity under different NPD and ECM arrival
rates. Three levels of NPD and ECM arrival rates are
combined in pairs according to their value. That is,
high NPD arrival rate is studied with high ECM
arrival rate, and low NPD arrival rate with low ECM
arrival rate.
There are altogether six sets of model variables,
and each of them has two or three possible values,
which is summarized in Appendix B. We run the
ICEIS 2009 - International Conference on Enterprise Information Systems
16
model 972 times altogether with the help of a
separate application of Arena call process analyzer
(PAN).
A partial results are presented in this paper due
to space limitation. The following two charts show
the impacts of overlapping, NPD departmental
interaction, and ECM effort on NPD Total Time and
Productivity under resource constraint of 60 units
from each department.
Figure 1: Simulation Results.
Based on the results we obtained, several
observations are made and its possible explanations
are given:
Observation Explanation
When there are unlimited
resources (200
resources/dept. in this case)
for NPD and ECM
activities, higher degree of
overlapping results in the
reduction of NPD lead time.
With more amount of
overlapping, there will be
more product development
activities executed before the
completion of the previous
ones. So products are
developed faster if there are
enough resources available.
When there are not enough
resources (60
resources/dept. in this case)
for NPD and ECM
activities, overlapping as
much as possible is no
longer recommended.
If only limited resources are
given, a medium level of
overlapping and high
departmental interaction
yields the optimal NPD lead
time. Firms need to make
compromise between shorter
value-added time but longer
wait time to grab resources
under higher degree of
overlapping.
As the Number of Resources This phenomenon is pretty
decreases, Productivity of
both NPD and ECM drops
off, but NPD with a higher
rate.
straightforward. When there
are fewer resources available,
the resource utilization raises,
sometimes even exceeds
100%. Then fewer NPDs and
ECMs will get adequate
resources to be completed in a
certain time period, runtime in
this case.
As the Number of Resources
decreases, Lead Time of
both NPD and ECM goes
up.
Even for those NPDs and
ECMs that get required
resources to be processed, the
total time (time an entity
enters the system until it exits)
will be longer due to longer
wait time for fewer resources
that are available.
A high departmental
interaction level results in
higher productivity and
shorter lead time than a low
departmental interaction
level, especially when
resources are limited.
Because each incoming ECM
may consume resources from
the three departments with
equal chance. With a total
resource demand unchanged,
if there is more departmental
interaction, there will be more
spare resources for the major
department to execute.
The Priority assigned to
NPD and ECM matters only
when the resources are
limited and the organization
choose to pursue a low level
of departmental interaction
(60-20-20 in this paper).
When high priority is
assigned to NPD,
productivity of NPD is
about 50% higher than the
situation in which high
priority is given to ECM,
while the productivity of
ECM is just slightly lower.
But at the same time, both
NPD and ECM take longer
to complete.
By assigning higher priority to
NPD, there are more NPD
entities coming out of the
system without affecting ECM
productivity much. However,
the price to pay is the longer
lead time for both NPD and
ECM since there are more
resource demands thus
resulting in a higher overall
resource utilization.
Organizations face tradeoffs
between productivity and lead
time in this situation.
The ECM Effort is not the
key factor of NPD/ECM
Productivity.
It affects NPD/ECM lead
time only when the
resources are limited and the
organization choose to
pursue a high level of
departmental interaction
(40-30-30 in this paper).
Recall that high level of
departmental interaction
means that minor departments
participate more while major
department allocates fewer
resources in its own
specialization phase.
So if an ECM is complex and
requires greater effort (10
resources from each
department in this case),
minor departments are much
easier to be out of resources
than low departmental
interaction case.
A SIMULATION MODEL FOR MANAGING ENGINEERING CHANGES ALONG WITH NEW PRODUCT
DEVELOPMENT
17
5 CONCLUSIONS
The NPD and ECM model framework introduced
above address several issues that earlier models
didn't. In this model, we capture important new
product design and development characteristics such
as iteration and overlapping of NPD process,
interaction among different functional areas,
resource constraints and its using priority. We also
take into account the size of NPD projects and ECRs
in terms of their arrival rates and processing effort.
From the simulation results, a number of
conclusions can be drawn:
1) ECM is an important aspect to the success
of an NPD project. On one hand, it solves
safety or critical functionality problems of a
product. And it reflects customer
requirements or technology developments.
On the other hand, it also consumes a
considerable amount of product
development resources which in turns
affects the lead time and productivity of
regular NPD activities significantly.
2) While each of the six model variables,
overlapping, NPD departmental interaction,
ECM effort, resource constraints, arrival
rate, and resource using priority, affects the
overall lead time and productivity of both
NPD and ECM by some extent, the effect
of resource constraints is most significant.
3) As stated in Section 4, this model addresses
decision-making suggestions for firms
under different organization environment
and resource constraint condition.
Specifically, when the resource capacity is
limited, a medium level of overlapping and
high departmental interaction is suggested
to optimize system resource utilization.
However, there are several aspects of this model
that need further investigation. First, the assumption
that one EC is confined in one NPD activity is not
always true. An EC that requires rework in a design
activity may propagate to other activities in design
or production phase. Future study should include
engineering change propagation as one feature of the
ECM process. Second, in the current model, we
assign to an NPD entity probabilities for feedback
iterations. However, when a new product project
needs to go back to earlier NPD activities for a
rework, subsequent activities need to be followed
again no matter how many times these activities are
repeated. In other words, an NPD entity has to go
through again all the downstream activities after
being sent back to the iteration starting point. Feed-
forward flexibility and learning effects for iteration
need to be considered in future work. Third, in this
model, it is assumed that NPD and ECM share the
same pool of resources with using priority. We could
let NPD and ECM have their own dedicated
resources. Or, NPD and ECM still use the same pool
of resources. But ECM requests for outsourcing
when resources are not available. In this case,
different utility costs can be set for using resources
within a department, cross departments, and for
outsourcing. Fourth, besides lead time and
productivity, other critical criteria such as resource
utilization, total cost, and customer satisfaction, can
be adopted to review and evaluate the impact of
ECM throughout NPD process.
REFERENCES
Bhuiyan, N., D. Gerwin, V. Thomson, 2004. Simulation of
the New Product Development Process for
Performance Improvement. Management Science.
50(12): 1690-1703.
Bhuiyan, N., G. Gr, V. Thomson, 2006. Engineering
Change Request Management in a New Product
Development Process. European Journal of Innovation
Management. 9(1): 5-19.
Browning, T. R., R. V. Ramasesh, 2007. A Survey of
Activity Network-Based Process Models for
Managing Product Development Projects. J. Product
Innovation Management. 16(2): 160-172.
Krishnan, V., S. D. Eppinger, D. E. Whitney, 1997. A
Model-Based Framework to Overlap Product
Development Activities. Management Science. 43(4):
437-451.
Loch, C. H., C. Terwiesch, 1999. Accelerating the Process
of Engineering Change Orders: Capacity and
Congestion Effects. J. Product Innovation
Management. 16(2): 145-159.
Moon, Y.B. 2007, Enterprise Resource Planning (ERP): a
review of the literature. International Journal of
Management and Enterprise Development, 4(3): 235-
264.
Terwiesch, C., C. H. Loch, 1999. Managing the Process of
Engineering Change Orders: The Case of the Climate
Control System in Automobile Development. J.
Product Innovation Management. 16(2): 160-172.
Wright, I.C., 1997. A Review of Research into
Engineering Change Management: Implications for
Product Design. Design Studies. 18: 33-42.
Kelton, W. David, R. P. Sadowski, D. T. Sturrock, 2006.
Simulation with Arena, The McGraw-Hill Companies.
London, 4
th
edition.
ICEIS 2009 - International Conference on Enterprise Information Systems
18